Every Stanford student is given a computer account which allows them to store and manipulate files using Stanford’s computers. After inspecting this system, known as AFS, I suspected it might be possible for anyone with an account to do two things:
1. Rapidly harvest the Stanford email address of every single student (and many alumni), enabling them to potentially spam tens of thousands of people.
2. Read the filenames—for example, “meganfox.jpg”—that every student had stored in the home directories of their Stanford accounts (unless the students had know-how to explicitly make those filenames private; I describe how to do this at the end of the article).
While for obvious reasons I’ll skip the technical details of how one could do this, I wrote to Stanford’s IT services (ITS) and they confirmed this was the case. This is not ITS’s fault: they didn’t design AFS (Carnegie Mellon did) and fixing an established system is hard. When I wrote to ITS, they responded rapidly and professionally, convincing me that there was no easy fix.
I would nonetheless suggest that this system is worth modifying because both the email addresses and the filenames invite abuses.
Firstly, people have attempted to harvest student emails in the past, and unscrupulous ASSU candidates, startup founders or anyone who’s ever used the phrase “FORWARD WIDELY” might do it in the future. One could also sell the addresses to a spam list. (Stanford does use an anti-spam system and other email controls to mitigate this problem.)
Secondly, being able to access filenames allows you, in many cases, to determine what classes a student is taking, and this could make a large-scale demographic analysis possible. Do students with female-sounding IDs take different classes than students with male-sounding IDs? What about students whose ethnic backgrounds are identifiable from their IDs? This isn’t the sort of analysis that Stanford wants to allow—I know because ITS got annoyed when I asked.
More pernicious are the potential dangers posed to individual students. One could investigate whether specific athletes are taking different classes from the general population. One could also look at whether children of famous parents take different classes, a question that those who scrutinize legacy admissions would be interested in.
It also would be easy to search for embarrassing filenames: I’d be willing to bet a lot of money that one of the many thousands of AFS users is storing porn. Even students who aren’t that technically sophisticated could investigate individual acquaintances—do you really want your embittered ex going through your computer? While most students probably save their most private files on laptops, should students who can’t afford laptops have a lesser expectation of privacy?
And beyond schoolwork and porn, there may be a lot of confidential, high-value data stored on AFS—unpublished research, ideas for companies and so on. Having worked at tech companies, I can attest that even a quick scan of a company’s filenames gives a lot of clues as to the directions it’s exploring.
I strongly advise against attempting to do any of this. Ignoring the fact that it’s a massive violation of trust, Stanford monitors AFS usage and ITS has made it clear that they will pursue offenders. But enforcement will naturally be imperfect, since it may be hard to discriminate between valid cluster usage and large-scale harvesting.
Ideally, Stanford students and affiliates should know that their filenames and emails cannot be harvested by one of the thousands of other users on the system and not have to trust the goodwill of others or retroactive sanctions. While this problem is not easy to fix, it is also not unfixable: The NSA, for example, probably doesn’t allow all its employees to access all its filenames. While it is true that in some cases we relax privacy for convenience, it’s unclear what legitimate reason someone would have for reading other students’ filenames if they cannot open the files themselves.
So I am writing this article for two reasons: First, to make you aware of the problem so you can protect yourself. There is no way to prevent your email address from being harvested, but you can make your filenames private quite easily:
If you have a Mac:
1. Click on the “Terminal” application. This might be a black square at the bottom of the screen. If you can’t find it there, you can click on the magnifying glass in the top right corner of the screen and then type “Terminal” to search for Terminal.
2. A little box should pop up. Click on it and log into your Stanford AFS account by typing “ssh email@example.com” where you replace “username” with your SUID. Press enter. You will have to enter your password.
3. Once you have logged in, type in “fsr setacl ~ system:campushosts none” and press enter. Your files are now hidden from other students.
If you do not have a Mac:
1. Go to a computer in one of the Stanford labs and log on using the Mac operating system; follow the instructions above.
You can find a more detailed explanation of what these commands do on the ITS website. If this sounds too complicated, there are a number of other fixes: your AFS account also gives you access to a “private” folder, and anything placed in there will be hidden from other students. Alternately, you can store files on Google Docs or Stanford Box, neither of which can be seen by other students. Or you could give all your files very boring names (“not_kitty_cat_porn.png”).
The second reason I am writing this article is my faith in the technical know-how of the Stanford community: I believe this system can be improved, and hopefully one of you will, too. Looking for a summer project?
Emma Pierson ’13
Emma graduated from Stanford last June with a B.S. in physics and a M.S. in computer science. A Rhodes Scholar, she plans to pursue a doctorate in computational biology at Oxford. In her free time, Emma writes for her blog Obsession with Regression.